Insomnia is associated with adverse outcomes in women in the perinatal period; thus, the assessment of insomnia is important for pregnant women. The Insomnia Severity Index (ISI) is an instrument used globally to assess the severity of insomnia. However, its factor structure and structural invariance for pregnant women have not been studied. Therefore, we aimed to conduct factor analyses to search for the best model to fit its structural invariance. A cross-sectional study with the ISI was conducted at one hospital and five clinics in Japan from January 2017 to May 2019. A set of questionnaires was administered on two occasions with a one-week interval. The study included 382 pregnant women ranging in gestational age from 10 to 13 weeks. One week later, 129 participants answered the retest. After exploratory and confirmatory factor analyses, the measurement and structural invariance between parity and two time points was tested. The two-factor structure model showed an acceptable fit for the ISI in pregnant women (χ2 (12) = 28.516, CFI = 0.971, RMSEA = 0.089). The model also showed satisfactory measurement and structure invariance between parity and time points. The findings indicate that the ISI's use would be appropriate for pregnant women as a two-factor subscale of "severity" and "impact", regardless of the parity or time point. The ISI's factor structure may vary by subject; hence, it is necessary to confirm the measurement and structural invariance of the subject for whom the ISI will be used. Furthermore, interventions that focus not only on total scores and cutoff points but also on the phenomenon of subscales should be considered.
Keywords: factor structure; insomnia severity index; measurement and structural invariance; parity; pregnancy; pregnant women.